Feature Extraction Algorithm Research Articles

A Method of EEG Signal Feature Extraction Based on Hybrid DWT and EMD

The processing and recognition of electroencephalogram (EEG) signal is the most important part of brain-computer interface (BCI) system, and the quality of signal processing and recognition is directly related to the effectiveness of BCI system. Aiming at the problems of incomplete removal of artifacts and inadequate retention of active components in EEG signal, a fusion method of wavelet transform (WT) and Fast Independent Component Analysis (FastICA) is utilized to preprocess the raw EEG signals. The fusion method can remove noise artifacts while preserving effective information. Aiming at the problems of poor time-frequency resolution and low classification accuracy of the traditional feature extraction method, a feature extraction algorithm on the basis of hybrid Discrete Wavelet Transform (DWT) and Empirical Mode Decomposition (EMD) is proposed. Firstly, DWT is used to analyze the pre-processed EEG signal to obtain a series of sub-band signals. Then, EMD decomposition is applied to subband signal and eigenmode function is extracted to complete feature integration. Finally, the feature extraction results are input into the Support Vector Machine (SVM) for classification. Comparative experiments show that the classification accuracy of the proposed method reaches 91.32%, which is significantly higher than other algorithms.

Feature extraction and fusion algorithm for infrared visible light images based on residual and generative adversarial network

With the application and popularization of depth cameras, image fusion techniques based on infrared and visible light are increasingly used in various fields. Object detection and robot navigation impose more stringent requirements on the texture details and image quality of fused images. Existing residual network, attention mechanisms, and generative adversarial network are ineffective in dealing with the image fusion problem because of insufficient detail feature extraction and non-conformity to the human visual perception system during the fusion of infrared and visible light images. Our newly developed RGFusion network relies on a two-channel attentional mechanism, a residual network, and a generative adversarial network that introduces two new components: a high-precision image feature extractor and an efficient multi-stage training strategy. The network is preprocessed by a high-dimensional mapping and the complex feature extractor is processed through a sophisticated two-stage image fusion process to obtain feature structures with multiple features, resulting in high-quality fused images rich in detailed features. Extensive experiments on public datasets validate this fusion approach, and RGFusion is at the forefront of SD metrics for EN and SF, reaching 7.366, 13.322, and 49.281 on the TNO dataset and 7.276, 19.171, and 53.777 on the RoadScene dataset, respectively.

Rapid Micro-Motion Feature Extraction of Multiple Space Targets Based on Improved IRT

Micro-motion feature extraction is of great significance for target recognition. However, traditional methods mostly focus on single target and struggle to correctly separate the severely overlapping micro-motion curves of multiple targets. In this paper, a rapid micro-motion feature extraction algorithm of multiple space targets based on inverse radon transform (IRT) with a modified model is proposed. First, the high-resolution range profile (HRRP) generated from echo is subject to binarization to improve the unstable estimation caused by noise. Then, the micro-motion period in a complicated multi-target scenario is obtained by a period estimation method based on the autocorrelation coefficients of binarized HRRP. To further improve the extraction accuracy, the IRT model of the micro-range curve is modified from the sine function to second-order sine function. By searching for the remaining unknown parameters in the model in conjunction with the period, the precise micro-range curves are quickly separated. Each time the curves of a target are extracted, they are removed, and the next extraction is carried out until all the targets have been searched. Finally, simulation and experimental results indicate that the proposed algorithm can not only correctly separate the micro-motion feature curves of multiple space targets under low signal-to-noise ratio (SNR) conditions but also significantly outperforms the original IRT in terms of extraction speed.

User Profile Construction Based on High-Dimensional Features Extracted by Stacking Ensemble Learning

Online social networks, as platforms for personal expression, have evolved into complex networks integrating political and social dimensions. This evolution has shifted the focus of network governance from addressing hacking activities to mitigating unpredictable social behaviors, such as the malicious manipulation of public opinion, the doxing of ordinary users, and cyberbullying. However, the sparsity of data and the concealed nature of user behavior pose significant challenges to existing network reconnaissance technologies. In this study, we focus on constructing user profiles on online social network platforms by extracting features to build deep user profiles based on behavioral patterns. Drawing inspiration from the 5Cs principle of credit evaluation, we refine it into a 3Cs principle tailored for user profiling on social network platforms and associate it with user behavioral patterns. To further analyze user behavior, a high-dimensional feature extraction method is proposed using an improved stacking ensemble learning model. Based on experimental data analysis, the most suitable base algorithms for high-dimensional feature extraction are identified. Experimental results demonstrate that the integration of high-dimensional features improved the behavior prediction accuracy of the profiling model by 9.26% on balanced datasets and enhanced the AUC (area under the curve) metric by 3.69% on imbalanced datasets. The proposed method effectively increases the depth and generalization performance of user profiling.

Enhancing Laser-Induced Breakdown Spectroscopy Spectral Quantification Through Minimum Redundancy and Maximum Relevance-Based Feature Selection

Laser-induced breakdown spectroscopy (LIBS) is a rapid, non-contact analytical technique that is widely applied in various fields. However, the high dimensionality and information redundancy of LIBS spectral data present challenges for effective model development. This study aims to assess the effectiveness of the minimum redundancy and maximum relevance (mRMR) method for feature selection in LIBS spectral data and to explore its adaptability across different predictive modeling approaches. Using the ChemCam LIBS dataset, we constructed predictive models with four quantitative methods: random forest (RF), support vector regression (SVR), back propagation neural network (BPNN), and partial least squares regression (PLSR). We compared the performance of mRMR-based feature selection with that of full-spectrum data and three other feature selection methods: competitive adaptive re-weighted sampling (CARS), Regressional ReliefF (RReliefF), and neighborhood component analysis (NCA). Our results demonstrate that the mRMR method significantly reduces the number of selected features while improving model performance. This study validates the effectiveness of the mRMR algorithm for LIBS feature extraction and highlights the potential of feature selection techniques to enhance predictive accuracy. The findings provide a valuable strategy for feature selection in LIBS data analysis and offer significant implications for the practical application of LIBS in predicting elemental content in geological samples.

Indian Sign Language Detection

- The communication gap remains one of the most significant barriers between individuals with hearing and speech impairments and the broader society. This project addresses this challenge by developing a real-time Indian Sign Language (ISL) detection system that leverages computer vision and machine learning techniques. By capturing hand gestures from video input, the system translates these movements into text or speech, enabling effective communication between ISL users and those unfamiliar with the language. Additionally, the system incorporates text-to-speech functionality, ensuring a seamless and humanized interaction experience. The proposed model utilizes Convolutional Neural Networks (CNNs) for image processing and gesture recognition, trained on a comprehensive dataset of ISL gestures. The framework employs preprocessing, feature extraction, and classification algorithms to accurately identify static and dynamic gestures. The system is designed to focus on the nuances of ISL, providing accurate recognition of gestures in real time while offering multilingual support. This initiative aspires to create an inclusive environment by empowering the hearing-impaired community and promoting better integration within society. By using cost-effective techniques, the project ensures scalability and practicality for everyday applications, making communication more efficient and inclusive. Keywords: Indian Sign Language (ISL), Gesture Recognition, Convolutional Neural Networks (CNNs), Real-time Communication

Non-Destructive Detection of pH Value During Secondary Fermentation of Maize Silage Using Colorimetric Sensor Array Combined with Hyperspectral Imaging Technology

The pH value of maize silage can accurately reflect its quality. In this study, a colorimetric sensor array (CSA) combined with hyperspectral imaging (HSI) was used to predict the pH value of maize silage during secondary fermentation. Seventeen color-sensitive dyes were used to construct the CSA, which was subsequently applied to capture the volatile odor profiles of maize silage samples. Hyperspectral images of the color-sensitive dyes on the CSA were acquired using the HSI technique. Different algorithms were used to preprocess the raw spectral data of each dye, and a partial least squares regression (PLSR) model was built for each dye separately. Subsequently, the adaptive bacterial foraging optimization (ABFO) algorithm was employed to identify three color-sensitive dyes that demonstrated heightened sensitivity to pH variations in maize silage. This study further compared the capabilities of individual dyes, as well as their combinations, in predicting the pH value of maize silage. Additionally, a novel feature wavelength extraction method based on the ABFO algorithm was proposed, which was then compared with two traditional feature extraction algorithms. These methods were combined with PLSR and backpropagation neural network (BPNN) algorithms to construct a quantitative prediction model for the pH value of maize silage. The results show that the quantitative prediction model constructed based on three dyes was more accurate than that constructed based on an individual dye. Among them, the ABFO-BPNN model constructed on the basis of combined dyes had the best prediction performance, with prediction correlation coefficient (RP2), root mean square error of the prediction set (RMSEP), and ratio of performance deviation (RPD) values of 0.9348, 0.3976, and 3.9695, respectively. The aim of this study was to develop a reliable evaluation model to achieve fast and accurate predictions of silage pH.

A Computationally Efficient Method for the Diagnosis of Defects in Rolling Bearings Based on Linear Predictive Coding

Monitoring the condition of rolling bearings is a crucial task in many industries. An efficient tool for diagnosing bearing defects is necessary since they can lead to complete machine failure and significant economic losses. Traditional diagnosis solutions often rely on a complex artificial feature extraction process that is time-consuming, computationally expensive, and too complex to deploy in practice. In actual working conditions, however, the amount of labeled fault data available is relatively small, so a deep learning model with good generalization and high accuracy is difficult to train. This paper proposes a solution that uses a simple feedforward artificial neural network (NN) for classification and adopts the linear predictive coding (LPC) algorithm for feature extraction. The LPC algorithm finds several coefficients for a given signal segment containing information about the signal spectrum, which is sufficient for further classification. The LPC-NN solution was tested on the Case Western Reserve University (CWRU) and South Ural State University (SUSU) datasets. The results demonstrated that, in most cases, LPC-NN yielded an accuracy of 100%. The proposed method achieves higher diagnostic accuracy and stability to load changes than other advanced techniques, has a significantly improved time performance, and is conducive to real-time industrial fault diagnosis.

Rapid Textile Fibers Classification Technology Based on Near-Infrared Spectroscopy and Machine Learning

The accurate identification of textile materials plays a crucial role in production control, quality inspection, and market regulation. In recent years, near-infrared (NIR) spectroscopy has been widely applied in the identification and classification of textile fibers due to its rapid, non-destructive, and efficient characteristics. Therefore, this study utilizes a handheld near-infrared spectrometer and chemometrics to provide a convenient, efficient, non-destructive, and green analytical method for the qualitative analysis of textiles. The article collects spectral data of fiber samples using a handheld NIR spectrometer and analyzes them with chemometric methods. Data preprocessing techniques, feature extraction algorithms, comprehensive sampling methods, and classification algorithms are employed to establish classification models. After ten-fold cross-validation, the optimal SG_RF model is obtained, with a classification accuracy of 93.7% on the test set and an AUC_macro of 0.928, demonstrating excellent performance. The article validates the feasibility of using near-infrared spectroscopy combined with chemometric methods for textile fiber classification, offering a rapid, non-destructive, and efficient approach for textile fiber classification.

Modified kernel global-local marginal fisher analysis for rolling bearing feature extraction

Abstract The vibration signals of rolling bearings usually exhibit high-dimensional, nonlinear, and non-Gaussian distribution characteristics due to long-term operation under complex working conditions. Therefore, we proposed a novel algorithm named Modified Kernel Global-Local Marginal Fisher Analysis (MKGLMFA) for bearing feature extraction and dimensionality reduction. The proposed MKGLMFA algorithm introduces the kernel function to map data into a high-dimensional space to represent data nonlinearly first. It enhances the within-class compactness and between-class dispersibility by considering spatial relationships and label information when constructing adjacency graphs and simultaneously exploits the local and global geometry of data. Furthermore, a bearing fault diagnosis approach is presented based on MKGLMFA. It first processes the original vibration signals through MKGLMFA to obtain low-dimensional manifold features. Then these characteristics were input into the K-nearest neighbor (KNN) classifier to achieve fault pattern recognition. The superiority of the proposed MKGLMFA algorithm in feature extraction is verified in comparison with some existing state-of-the-art machine learning methods on three rolling bearings datasets. And the subsequent classification diagnosis experiments indicate the effectiveness and high efficiency of the newly raised MKGLMFA algorithm. In comparison with the representative diagnosis methods, the proposed method can extract more sensitive discriminant features, and the classification accuracy of diagnosis is significantly improved in consequence.

Early Prediction of Cardio Vascular Disease (CVD) from Diabetic Retinopathy using improvised deep Belief Network (I-DBN) with Optimum feature selection technique

Cardio Vascular Disease (CVD) is one of the leading causes of mortality and it is estimated that 1 in 4 deaths happens due to it. The disease prevalence rate becomes higher since there is an inadequate system/model for predicting CVD at an earliest. Diabetic Retinopathy (DR) is a kind of eye disease was associated with increasing risk factors for all-causes of CVD events. The early diagnosis of DR plays a significant role in preventing CVD. However, there are many works have been carried out on classification of the disease but they focused less on feature selection and increasing the accuracy of the model. The proposed work introduces Improvised Deep Belief Network named I-DBN to resolve the above mentioned problems and mainly to concentrate on improving the entire performance of the model leading to the unbiased output. We used Principal Component Analysis (PCA) and Particle Swarm Optimization (PSO) algorithm for feature extraction and selection respectively. Five performance metrics have been used to assess the proposed model. The results of I-DBN outperform other state-of-the-art methods. The result validation ensures that I-DBN can deliver trustworthy recommendations to doctors to treat the patients by enhancing the accuracy of CVD prediction up to 98.95%.

Characterization of Spatial Cognitive EEG Signals Using Normalized Adjusted Permutation Conditional Mutual Information

Spatial cognitive ability, a fundamental domain within the human cognitive system, involves the perception, interpretation, and manipulation of spatial environments. This study introduces a new EEG feature extraction algorithm, Normalized Adjusted Permutation Conditional Mutual Information (NAPCMI), to improve the accuracy of spatial cognition assessments. By capturing the symmetry and temporal dependencies within EEG signals during spatial cognition tasks, NAPCMI enhances the ability to extract relevant features. The study validates NAPCMI using a BCI-VR spatial cognition assessment system, incorporating gesture recognition. Results demonstrate that NAPCMI outperforms traditional methods in feature extraction, highlighting its potential for advancing the understanding and assessment of spatial cognitive abilities. The findings also emphasize the significance of specific EEG frequency bands, such as Delta and Beta1, in spatial cognition tasks, further validating NAPCMI’s effectiveness.

Resolving tissue complexity by multimodal spatial omics modeling with MISO.

Spatial molecular profiling has provided biomedical researchers valuable opportunities to better understand the relationship between cellular localization and tissue function. Effectively modeling multimodal spatial omics data is crucial for understanding tissue complexity and underlying biology. Furthermore, improvements in spatial resolution have led to the advent of technologies that can generate spatial molecular data with subcellular resolution, requiring the development of computationally efficient methods that can handle the resulting large-scale datasets. MISO (MultI-modal Spatial Omics) is a versatile algorithm for feature extraction and clustering, capable of integrating multiple modalities from diverse spatial omics experiments with high spatial resolution. Its effectiveness is demonstrated across various datasets, encompassing gene expression, protein expression, epigenetics, metabolomics and tissue histology modalities. MISO outperforms existing methods in identifying biologically relevant spatial domains, representing a substantial advancement in multimodal spatial omics analysis. Moreover, MISO's computational efficiency ensures its scalability to handle large-scale datasets generated by subcellular resolution spatial omics technologies.

Anomaly detection in virtual machine logs against irrelevant attribute interference.

Virtual machine logs are generated in large quantities. Virtual machine logs may contain some abnormal logs that indicate security risks or system failures of the virtual machine platform. Therefore, using unsupervised anomaly detection methods to identify abnormal logs is a meaningful task. However, collecting accurate anomaly logs in the real world is often challenging, and there is inherent noise in the log information. Parsing logs and anomaly alerts can be time-consuming, making it important to improve their effectiveness and accuracy. To address these challenges, this paper proposes a method called LADSVM(Long Short-Term Memory + Autoencoder-Decoder + SVM). Firstly, the log parsing algorithm is used to parse the logs. Then, the feature extraction algorithm, which combines Long Short-Term Memory and Autoencoder-Decoder, is applied to extract features. Autoencoder-Decoder reduces the dimensionality of the data by mapping the high-dimensional input to a low-dimensional latent space. This helps eliminate redundant information and noise, extract key features, and increase robustness. Finally, the Support Vector Machine is utilized to detect different feature vector signals. Experimental results demonstrate that compared to traditional methods, this approach is capable of learning better features without any prior knowledge, while also exhibiting superior noise robustness and performance. The LADSVM approach excels at detecting anomalies in virtual machine logs characterized by strong sequential patterns and noise. However, its performance may vary when applied to disordered log data. This highlights the necessity of carefully selecting detection methods that align with the specific characteristics of different log data types.

GPR Signal Processing for Landmine Detection: A Comparative Study of Feature Extraction and Classification Algorithms

GPR Signal Processing for Landmine Detection: A Comparative Study of Feature Extraction and Classification Algorithms

Facial Feature Extraction Algorithm Based on Improved YOLOv7-Tiny

Facial Feature Extraction Algorithm Based on Improved YOLOv7-Tiny

Fault detection for industrial processes based on time-serial maximal deviation analysis

Abstract The application of feature extraction algorithms in data-driven fault detection has been widely investigated for industrial processes, with a wide range of multivariate analytical methods available in the literature for this purpose. However, mainstream methods all focus on exploiting the systematic variation in historical normal samples and a fault is usually isolated as a deviation from the characterized normal signatures. For the sole purpose of fault detection in dynamic processes, it may be more effective to timely and adaptively uncover deviations inherent in time-serial variations for online sequential samples of interest, rather than focusing solely on extracting latent representations from normal samples. Therefore, a novel feature extraction algorithm called time-serial maximal deviation analysis (TSMDA) is proposed. TSMDA is designed to derive a projecting scheme with dynamically determined projecting vectors in a timely manner, so that possible time-serial deviations in newly measured sequential samples can be adaptively uncovered to the greatest extent possible. Most importantly, TSMDA cannot be executed without the involvement of online monitored samples and it is expected to guarantee consistently enhanced fault detectability. The effectiveness of the proposed fault detection method is validated through practical experiments on two industrial processes and the influence of key model parameters on fault detectability is also evaluated.

Machine Learning Methods in IoT Based Embedded Systems for Classifying Physical Faults in Water Distribution Networks

Water is the most important factor for the survival of living things on Earth. Although 70% of the Earth is water, the amount of drinkable water is approximately 0.3%. Therefore, creating a sustainable water policy and carrying out studies are very important for our world and our future. Most of the potable water resources are physical losses. In the evaluations made based on metropolitan municipalities, it was seen that the water loss rate was approximately 50%. The study aims to find water pipe faults using IoT (Internet of Things) based machine learning classifiers to prevent physical losses in water distribution networks. Within the scope of this study, an experimental environment was created and an IMU (Inertial Measurement Unit) sensor was fixed on plastic pipes of different diameters and lengths. Vibration data collected in different scenarios (pressure, etc. factors) were transferred to the ThingSpeak platform over the internet. The transferred data could be monitored in real-time on a server. Physical damage in the pipes was detected using signal pre-processing, feature extraction, and feature selection algorithms on vibration data. In the study, damages were classified using machine learning-based classification (Decision Trees, k-Nearest Neighbors, Linear Discriminant, Support Vector Machines) methods to predict the type of damage (solid, hole, multi-hole). The data set revealed within the scope of the study is thought to lead to scientific studies in this field. The results obtained are close to the state-of-the-art results.

One Receptor Show: Evaluation of Feature Extraction Methods for a Single Gas Sensor Based on α-Tocopherol-Substituted Zinc Phthalocyanine.

A consistent part of gas sensor research activities aims to improve sensing performances by synthesizing new sensing materials, improving the selection of elements in arrays, and optimizing the feature extraction and classification algorithms. This paper combines most of these aspects to confer selectivity to a low-selectivity sensor by using feature extraction algorithms applied to the sensor response kinetics. Several algorithms were employed to represent the kinetic behavior of the sensor response during the adsorption and desorption phases. In detail, we utilized fitting functions, dynamic descriptors, and a mirrored fast Fourier transform (FFT) to extrapolate information from the desorption and adsorption phases. As sensitive material, we synthesized as a receptor a novel Zn(II) phthalocyanine derivative bearing long aliphatic chains, namely, 4-[2,5,7,8-tetramethyl-2-(4,8,12-trimethyl-tridecyl)-chroman-6-yloxy] units that allow good solubilization of the complex in most of the solvents tested and confer a large set of sites to potentially interact with volatile compound (VC) analytes by different portions of the molecular skeleton. After deposition onto a quartz crystal microbalance (QMB) transducer, this single sensor has been tested to recognize five VCs as chemical probes of different chemical families. Remarkably, the results disclose that responses at equilibrium are mainly correlated with nonspecific interactions, while the kinetic parameters are more correlated with stronger and more specific interactions. Finally, the desorption phases bear a higher information content about the chemical identities of the compounds. Eventually, features extracted by a modified version of the FFT obtained the highest clustering (principal component analysis) and classification performances (95% classification accuracy with linear discriminant analysis), suggesting the possibility of developing virtual electronic noses based on a single sensor and a minimal set of descriptors.

Transform invariant low rank texture feature extraction and restoration algorithms for architectural decoration surface patterns

Transform invariant low rank texture feature extraction and restoration algorithms for architectural decoration surface patterns